1. Field of the Invention
The present invention relates to a system for data management managing data of a relatively long length such as video data and audio data recorded on variable length recording areas of a randomly accessible recording medium such as a magnetooptic (MO) disc or hard disc (HD), more particularly relates to randomly accessible recording media, a method of data management, a method of generating and recording management data, and data management apparatuses used in such a system.
2. Description of the Related Art
When recording or reproducing data to or from a randomly accessible recording medium such as a magnetooptic disc or hard disc, it becomes necessary to manage data managing what data are recorded at what recording areas of the recording medium.
In the management of data in MS-DOS (trademark of Microsoft Corporation), the operating system of personal computers, as shown in FIG. 1A to FIG. 1C, use is made of the FAT system using a table called a "file allocation table" (FAT). In the management of data in the operating system of general purpose computers known as UNIX (trademark of ATT), as shown in FIG. 2A to FIG. 2C, use is made of the i-node system using a table called an "i-node".
FIG. 1A shows the content of a file entry, FIG. 1B shows the content of the FAT, and FIG. 1C shows the recording areas on a recording medium.
In MS-DOS, recording areas (file data areas) are allocated in units of clusters on the recording medium for each of the files containing data to be managed (recordal data). A file entry is prepared for each file as shown in FIG. 1A, and a column of the FAT (FAT entry) shown in FIG. 1B is prepared for each cluster.
As shown in FIG. 1A, a file entry has recorded in it the data showing the name of the file (file name) stored in the corresponding clusters and the no. of the cluster at which the head portion of the file is recorded (head FAT no.)
As shown in FIG. 1B, each FAT entry has recorded in it data showing the state of use of the cluster, that is, the state of non-use, two types of states of use, and a defective state. Among the two types of states of use, when the cluster is in use and a continuation of the recordal data is recorded in another cluster, data showing the next cluster no. is recorded, while when the cluster is in use and there is no continuation of the recordal data recorded in another cluster, an "end-of-file" (EOF) showing that the cluster is the last cluster is recorded.
Accordingly, in the clusters of the recording medium, as shown by the hatching in FIG. 1C, when the file "file 1" ("file 1" being a file name) is recorded at the clusters 3, 4, and 8, "file 1" is recorded as the data showing the file name of the file entry and "3" is recorded as the head FAT no.
Further, the FAT entry (FAT3) corresponding to the cluster 3 has recorded in it the no. "4" of the next cluster on which a continuation of the recordal data is recorded so as to show that a continuation of the recordal data is recorded at another cluster. Similarly, the FAT4 has the next cluster no. "8" recorded in it. The FAT8 has recorded in it the "end-of-file (EOF) for showing that the cluster is in use, but that no continuation of the recordal data is recorded at another cluster.
MS-DOS, when for example reading a file "file 1", first runs a search referring to the data showing the file names of the file entries so as to find the file entry corresponding to the file "file 1". Next, it refers to the head FAT no. of the file entry it found and then refers to the FAT 4 from the content of the FAT3. Finally, it recognizes the "end-of-file" of the FAT8 from the content of the FAT4 and stops referring to the FAT entries. In this way, MS-DOS successively traces back from the FAT entry shown by the head FAT no. to the FAT entry in which the "end-of-file" is recorded to thereby be able to read from the recording medium the file "file 1" recorded at the clusters 3, 4, and 8.
FIGS. 2A to 2C are views illustrating the management of data by the i-node system, wherein FIG. 2A shows the content of a file entry, FIG. 2B shows the content of an i-node, and FIG. 2C shows recording areas on a recording medium.
In the i-node system, a file entry shown in FIG. 2A corresponding to each file and an i-node shown in FIG. 2B are prepared.
In the i-node system, a file entry has recorded in it data showing the file name and an i-node no. showing the i-node of that file. An i-node has recorded in it the file size showing the length of data of the file and block pointers (direct assignment pointers) directly showing for example 1,024 byte unit recording areas (blocks) at which the file is recorded. Note that the i-node sometimes has recorded data showing attributes of the file. Further, it sometimes has recorded an indirect assignment pointer showing a block in which is recorded a plurality of direct assignment pointers instead of the direct assignment pointers.
For example, as shown in FIGS. 2B and 2C, when the content of the i-node is that the file "1file 2" has data of a length of 3000 bytes and is recorded at blocks 3, 4, and 8, UNIX refers to the i-node and can recognize that the data of the file "file 2" is recorded at all of the blocks 3 and 4 and at the portion of the block 8 from the head to the 952nd byte.
The above-explained file entry and FAT or i-node are all recorded at predetermined areas on the recording medium.
If data is managed by the above-explained FAT system and i-node system, one FAT entry or one direct pointer of the i-node becomes necessary for each unit of the recording area (the above-mentioned cluster or block). Therefore, when managing data for a file containing recordal data of extremely long individual lengths such as audio data or video data, there is the problem that the recording areas for recording the FATs or i-nodes in the recording medium become extremely large.
This problem will be explained in more detail using as an example the case of a length of data of a cluster or block of 1 kbyte and a length of data of the recordal data of 100 Mbytes.
When managing data of a file containing such recordal data by the PAT system, the number of FAT entries required becomes 100,000. Therefore, to identify each of the 100,000 PAT entries, 4 bytes of data become necessary. Therefore, it is learned that 400 kbytes of memory are necessary for storing the FAT entries.
When managing data of a file containing such recordal data by the i-node system, the number of direct assignment pointers necessary becomes 100,000. Therefore, to identify each of the 100,000 direct assignment pointers, 4 bytes of data become necessary to show a single FAT entry. Therefore, it is learned that 400 kbytes of memory are necessary for recording all of the direct pointers in the same way as the FAT system.
In the FAT system or i-node system, to reduce the number of the PAT entries or the number of the i-node direct pointers, it may be considered to increase the length of the data of the cluster or the block for example. However, in these systems, the data of the files is managed in units of clusters or blocks, so when the length of the recordal data is not a whole multiple of the length of data of the clusters or the blocks, there is the problem of a large amount of waste occurring in the cluster or block recording the file which contains the final portion of the recordal data.